| Methods for identifying neuronal spikes -> Monitor Keywords |
|
|
  Methods for identifying neuronal spikesUSPTO Application #: 20060167369Title: Methods for identifying neuronal spikes Abstract: A method for identifying neuronal spikes (extracellular action potentials) is described wherein measured microelectrode readings from tissue are reviewed to identify spikes (successive readings having prolonged rises and/or falls). The frequency of such spikes as a function of their amplitude assumes a bimodal distribution wherein higher amplitude spikes represent neuronal spikes (signal) and lower amplitude spikes represent noise, and thus the higher amplitude spikes can be assumed to be neuronal spikes. Neuronal spikes from the same neuron can then be assumed to have substantially the same waveform shape and period, with the only significant difference between them being the scaling of their amplitudes (i.e., the amplitudes of spikes from the same neuron tend to be proportionate at any given time along their period). Thus, by testing identified neuronal spikes for matching timing and for proportional amplitudes, the neuronal spikes may further be identified as coming from the same or different neurons. (end of abstract) Agent: Dewitt Ross & Stevens, S.c. US Bank Financial Centre - Madison, WI, US Inventors: Erwin B. Montgomery, He Huang, John T. Gale USPTO Applicaton #: 20060167369 - Class: 600544000 (USPTO) Related Patent Categories: Surgery, Diagnostic Testing, Detecting Brain Electric Signal The Patent Description & Claims data below is from USPTO Patent Application 20060167369. Brief Patent Description - Full Patent Description - Patent Application Claims
CROSS-REFERENCE TO RELATED APPLICATIONS [0001] This application claims priority under 35 USC .sctn.119(e) to U.S. Provisional Patent Application 60/638,554 filed 22 Dec. 2004, the entirety of which is incorporated by reference herein. FIELD OF THE INVENTION [0002] This document concerns an invention relating generally to methods for interpreting neurological signals, and more specifically to methods for discriminating and sorting signals generated by individual neurons. BACKGROUND OF THE INVENTION [0003] Deep brain stimulation (DBS) is a methodology wherein therapeutic electrodes are implanted within the brain to deliver timed impulses to desired nerve centers, and can be used to treat a variety of disorders, in particular movement disorders such as Parkinson's disease and dystonia. A major challenge with DBS relates to determination of where to place therapeutic electrodes: while medical imaging (e.g., magnetic resonance imaging) can provide a starting point for identification of placement locations, since the therapeutic electrodes must be precisely placed (e.g., in those regions of the brain giving rise to muscle tremors), it is necessary to obtain a more detailed map of brain structures. A common technique is to advance a needle-like probe (or multiple such probes) into the brain, with each probe bearing one or more reading microelectrodes. The reading microelectrodes measure neuronal activity, as indicated by extracellular action potentials, which are in essence voltage spikes caused by neuronal firings. As readings are taken, the patient--who is generally awake during the procedure--may be requested to perform some action (e.g., move an arm or leg), thereby provoking neuronal firings. By looking at the location of the probe (i.e., probe location and depth) and the characteristics of the measured neuronal spikes (e.g., shape, frequency, etc.), the regions of the brain traversed by the probe can be mapped: spike characteristics can be correlated with those known to exist in certain portions of the brain, changes in spike characteristics can indicate interfaces between different regions of the brain, and so forth. Additionally, stimulating input (voltage) pulses can simultaneously be delivered to regions of the brain which are candidates for electrode implantation to determine their physiological effect (e.g., whether tremors are reduced, whether the patient experiences some change in feeling, etc.), with such stimulating input pulses being delivered via an input point on the probe or via a separate electrode spaced from the probe. Once the map of the brain is generated and candidate locations for implantation of therapeutic electrodes are identified, therapeutic electrodes may be permanently implanted, with the therapeutic electrodes being connected to a power supply which delivers an input signal suitable to reduce or eliminate tremors, or to attain some other desired effect. [0004] However, the process of mapping the brain can be a difficult one. It can be extremely difficult to discern neuronal spikes from background noise (and from any stimulating input pulses) within these data, particularly owing to the wide variety of characteristics neuronal spikes may have; spikes can vary widely in their shape, amplitude, period, frequency, and so forth. The same neuron can even generate different spike readings over time, both owing to variability in the neuron itself and owing to factors such as nearby pulsing blood vessels creating small changes in neuron-to-probe spacing. Experienced neurologists and others can over time gain skill in identifying the neuronal spikes of individual neurons from probe readings, but since probes may contain large arrays of reading microelectrodes, thereby generating multiple streams of reading data, it is virtually impossible for human operators to successfully process all of the generated data. It would therefore be useful to have improved methods available for identifying neuronal spikes, in particular methods which require minimal human review and supervision, and which might therefore be suitable for use in expert systems and other automated or semi-automated systems for probe data review. SUMMARY OF THE INVENTION [0005] The invention, which is defined by the claims set forth at the end of this document, is directed to methods and systems which at least partially alleviate the aforementioned problems. A basic understanding of some of the preferred features of the invention can be attained from a review of the following brief summary, with more details being provided elsewhere in this document. [0006] After a series of microelectrode readings (i.e., sampled readings of amplitude/voltage over time) are obtained, the readings may (if desired) be cleansed of any artifacts arising from any stimulating input pulses that were delivered to the surrounding tissue, thereby allowing easier identification of any neuronal spikes. To remove artifacts, a detected artifact--whose timing can be relatively readily determined, either owing to the known timing of the stimulating input pulses and/or owing to the fact that stimulating input pulses tend to have higher amplitude than neuronal spikes--can be subtracted from any later artifacts detected in the readings, or an average of several artifacts can be subtracted from the artifacts in the microelectrode readings. However, it has been found that subtraction may not always efficiently remove artifacts from microelectrode readings, particularly where artifacts have high-frequency components. Thus, it is preferred that at high-frequency regions of artifacts in microelectrode data, the amplitudes of the microelectrode readings be "flattened"--that they simply be zeroed or otherwise attenuated. The remainders of the artifacts can then be removed from the microelectrode readings via subtraction. This process is exemplified, for example, by the accompanying FIGS. 4A and 4B, wherein FIG. 4A shows a series of raw or "uncleaned" microelectrode readings with plainly evident input pulse artifacts, and FIG. 4B shows the same series after flattening of the artifacts in the high-frequency regions denoted F and subtracting the artifacts in the low-frequency regions denoted S. [0007] The microelectrode readings may then be reviewed to identify series of successive readings having continuously increasing amplitude (as might occur at the outset of a positive-amplitude peak of a neuronal spike) and/or continuously decreasing amplitude (as might occur at the outset of a negative-amplitude valley of a neuronal spike), with each such series of readings representing a spike candidate. It has been found that the probability distribution of several successive amplitudes constantly increasing (or decreasing) as a function of amplitude (e.g., of the first amplitude in the series) is a bimodal one which exhibits two peaks, one being at a lower amplitude and having a higher proportion of spike candidates, and the other being at a higher amplitude and having a lower proportion of spike candidates. The latter of these peaks represents a signal peak populated by neuronal spikes, and the former of these represents a noise peak, i.e., it does not contain neuronal spikes. This is illustrated in the accompanying FIGS. 6A-6C, which show (at the right) the amplitude distribution of candidate spikes at a variety of signal-to-noise ratios (FIG. 6A having lowest noise and FIG. 6C having the highest), with exemplary sections of microelectrode readings at these signal-to-noise ratios being depicted at left. Thus, neuronal spikes can be chosen from the spike candidates by excluding spike candidates which are clustered about a lower amplitude, and/or which are clustered about the most commonly occurring amplitude (i.e., by excluding those spike candidates situated in the noise peak). Conversely, neuronal spikes can be chosen from the spike candidates by including spike candidates clustered about a higher amplitude, and/or which are clustered about the second most commonly occurring amplitude (i.e., by including those spike candidates situated in the signal peak). A preferred method of identifying the spike candidates representing neuronal spikes is to look to the amplitude at the center of the noise peak (this amplitude representing the median amplitude, or at least approximately so, of the candidate spikes in the noise distribution), and then double this value to set a threshold amplitude value defining the nominal right tail of the noise distribution (and thus the nominal left tail of the signal distribution). Thus, spike candidates having amplitudes greater than this threshold amplitude value (greater than twice the median amplitude of the noise distribution/peak) can be regarded to be neuronal spikes. [0008] In the foregoing method, it is preferred that the determination of whether increasing (or decreasing) slope is present, and thus whether microelectrode readings correspond to a spike candidate, be based on at least three points (i.e., three sampled microelectrode readings), and preferably no more than five points, at least when these points are chosen at a frequency of 25 kHz. Since the introductory upward and downward slopes of most neuronal spikes tend to be approximately 0.2 ms in duration or longer (see FIG. 9 for examples), it is preferred that slopes be tested by a series of points ranging over approximately 0.075-0.175 ms, or more preferably 0.1-0.15 ms. [0009] After neuronal spikes are identified, it is then useful to assign them specific starting and ending times, since this can ease spike grouping (as described below) and/or other spike analysis activities. A preferred nominal starting point for a neuronal spike is at a zero crossing (point of zero amplitude) prior to the positive threshold amplitude value (assuming the spike starts with a positive peak), or prior to the negative threshold amplitude value (assuming the spike starts with a negative peak). Similarly, a nominal ending time can be assigned by locating a zero crossing after the negative or positive threshold amplitude value on the following negative or positive peak. Most preferably, the second zero crossings prior to and after the threshold values are chosen for use as the start and end of a neuronal spike, since these tend to adequately capture neuronal spikes without including excessive noise at the start and/or end of the spikes. [0010] If desired, spikes may then be grouped by neuron so that one may determine how many neurons are present in a set of microelectrode readings, and may review the spike characteristics of each neuron. A preferred grouping method is to review all identified neuronal spikes in sequence, and with each one, compare its characteristics to those of "template" spikes (if any) stored in a database/library. If the spike meets the characteristics of the spike template to within a predefined tolerance, it is grouped with the spike template; if it does not, it is stored as a template against which later spikes may be compared. Preferably, the characteristics used for comparison between spikes and templates are time scaling (i.e., whether at least substantially identical time exists between the amplitude maxima and amplitude minima of the spike and the template against which it is being compared), and amplitude proportionality (i.e., whether the maxima and minima, or other specific points along the spike, are all proportionate to the corresponding values on the template to which the spike is compared). Other characteristics of spikes could also be used for comparison and grouping, e.g., root mean square amplitude of spikes, the differences between amplitude maxima and minima, etc. [0011] Since this template-matching methodology effectively adopts the first-encountered member of any distinct group of spikes as a template for that group, the spike adopted as the template for the group may not necessarily represent the average spike within the group. Thus, it is also useful to store a running average of the spikes in a group so that this average can be used as a representative member of the group if/when needed. [0012] Grouping, whether performed by template matching or by other methods, can also effectively assist with spike discrimination and artifact cleansing. For example, if groups of spikes include only a single or few neuronal spikes therein, this may indicate that the spikes in the group are not in fact neuronal spikes, but are rather noise spikes, and thus these may be excluded from the template/neuronal spike database. Conversely, if groups of spikes contain very large numbers of spikes, this may also indicate the presence of non-neuronal spikes which should be excluded--for example, if artifact cleansing is not performed, a very large group of spikes may arise wherein these spikes are in fact artifacts. The foregoing methods therefore provide a means for identifying neuronal spikes and sorting them according to their source (neuron) in a manner which needs no human intervention/supervision, and which can readily be implemented in computers, medical instrumentation, expert systems, and other devices for automatic (or substantially so) analysis of microelectrode readings. Further, the methods are computationally simple, and can be run in real time as microelectrode readings are collected, in addition to (or instead of) in stored collections of previously-obtained microelectrode readings. The methods are also well suited for the processing of very large amounts of microelectrode readings, such as those that are provided by probes with dense arrays of multiple microelectrodes thereon, and for which the collected data becomes so immense that it is practically unprocessable by systems requiring human supervision. Further advantages, features, and objects of the invention will be apparent from the following detailed description of the invention in conjunction with the associated drawings. BRIEF DESCRIPTION OF THE DRAWINGS [0013] FIG. 1 depicts an exemplary plot of microelectrode data (voltage versus time) as measured by a probe microelectrode within a brain, with the solid arrows indicating the time of stimulating input pulses delivered to the brain (and with the artifacts of these pulses, i.e., their measured output, being visible as large-amplitude spikes), and with the open arrows indicating the times of neuronal spikes. [0014] FIG. 2A depicts the artifact of a stimulating input pulse delivered to the brain over a timescale which is expanded in comparison to the timescale of FIG. 1. [0015] FIG. 2B depicts several stimulating input pulse artifacts from the same measurement session (as in FIG. 1) with the waveforms of all input pulse artifacts time-shifted to begin at the same time, showing that input pulse artifacts from the same measurement session can be substantially identical, and thus can be characterized to allow their removal from the measured microelectrode data. [0016] FIG. 3 schematically depicts a scheme for removal of input pulse artifacts from measured microelectrode data wherein measurements of the autopower are generated starting near the time of the delivery of the stimulating input pulse. Peaks in the autopower of the measured microelectrode readings (including the input pulse artifact) are then used to define a flattening region F wherein the microelectrode readings may be flattened (artificially attenuated), and also a subtraction region S wherein the values of a representative artifact may be subtracted from the measured microelectrode readings, with such flattening and subtraction resulting in removal of the artifact from the measured microelectrode readings. [0017] In FIGS. 4A and 4B, FIG. 4A illustrates raw data (original measured microelectrode readings) with input pulse artifacts, and FIG. 4B then illustrates the same data after the artifacts are cleansed therefrom by use of the foregoing flattening and subtraction method (with flattening regions F and subtraction regions S again being shown). [0018] FIG. 5 schematically depicts a method for discrimination of neuronal spikes from background noise within measured microelectrode readings, wherein a series of successive sampled microelectrode readings (here every other one of the sampled microelectrode readings) is tested to see whether a prolonged increase in amplitude occurs, thereby indicating the possible presence of a neuronal spike. [0019] FIGS. 6A-6C then illustrate a series of measured microelectrode readings centered about a neuronal spike at a variety of signal-to-noise levels (with noise increasing from FIG. 6A-6C). An adjacent histogram shows the frequency of occurrence of a prolonged amplitude increase (such an amplitude increase being indicative of a "candidate neuronal spike") versus the amplitude of the candidate neuronal spikes, with such histograms showing a bimodal distribution wherein the peak at low amplitude is indicative of noise spikes and the peak (or "hump") at higher amplitude is indicative of neuronal spikes, and wherein the vertical lines in the histograms--set at twice the amplitudes of the noise peak--is set as a threshold above which neuronal spikes are assumed to occur. (Note that the plots of the measured microelectrode readings only represent a small portion of the data used to generate the histograms.) Continue reading... Full patent description for Methods for identifying neuronal spikes Brief Patent Description - Full Patent Description - Patent Application Claims Click on the above for other options relating to this Methods for identifying neuronal spikes patent application. ###  How KEYWORD MONITOR works... a FREE service from FreshPatents1. Sign up (takes 30 seconds). 2. Fill in the keywords to be monitored. 3. Each week you receive an email with patent applications related to your keywords. Start now! - Receive info on patent apps like Methods for identifying neuronal spikes or other areas of interest. ### Previous Patent Application: Method and apparatus for providing improved assessment of a physiological condition of a patient Next Patent Application: System and method for prediction of adverse events during treatment of psychological and neurological disorders Industry Class: Surgery ### FreshPatents.com Support Thank you for viewing the Methods for identifying neuronal spikes patent info. IP-related news and info Results in 2.08136 seconds Other interesting Feshpatents.com categories: Computers: Graphics , I/O , Processors , Dyn. Storage , Static Storage , Printers |
||
